WCDMA Network Performance

Optimisation Based on Analysis

of Network Statistics

Ari Lehtinen
Supervisor: Professor Riku Jäntti

© Omnitele Ltd. 2007 1

 Contents

• Background
• Research problem
• Research methods
• Optimisation process
• Measuring network performance
• Results
• Case study
• Conclusions
• Future research

© Omnitele Ltd. 2007 2

 Background

• WCDMA network performance needs to be monitored and

optimised as traffic in networks is increasing
• Network monitoring is based on correctly defined KPIs
• KPI definitions for WCDMA networks are immature
• Network statistics (KPIs) are needed in network performance
optimisation

© Omnitele Ltd. 2007 3

 Research problem

How to monitor WCDMA network performance and how to

use the monitoring information to optimise the network
performance?

Objectives:

• Define the KPIs that measure the radio network

performance in WCDMA networks
• Examine how to use the defined KPIs in radio
network performance optimisation

© Omnitele Ltd. 2007 4

 Research methods

• Literature study
– Various technical specifications
– Several books and publications

• Interviews
– Radio network planners
– Radio network experts

• Case study
– Optimisation of an operator’s live network

© Omnitele Ltd. 2007 5

 Optimisation process

• Reasons that lead to optimisation :

– Improve the performance
– Business reasons (cost-effective)
– Troubleshooting

© Omnitele Ltd. 2007 6

 Measuring network performance

• Network statistics
– Key Performance Indicators (KPI)
• Field measurements
– Drive tests
– Interface probes
• End user feedback

© Omnitele Ltd. 2007 7

 Network statistics

• Network statistics are collected from different network elements

with counters
• Different types of counters are used
• KPIs are needed to provide information of the network
performance
• Raw counter data too detailed to be used in monitoring and
optimisation (Some counters can be used as KPIs)

© Omnitele Ltd. 2007 8

 Defining the KPIs

• KPIs are composed from several counters

• KPI categories
– Accessibility
– Retainability
– Integrity
• Documentation of KPIs is important
– Same KPI can be defined from different counters or formula can be
incorrect
• Measurement period must be reasonable
– Too much averaging if too long
– Not enough statistical information if too short

© Omnitele Ltd. 2007 9

 Results
Category
• A set of KPIs
Accessibility
• Optimisation based on KPIs:
– Optimisation is performed for
each category
– Find the worst performing cells
Retainability
– Find the reasons behind the poor
performance
– Make the changes in the network
– Monitor the performance after the
changes
• A case study was conducted in
order to verify the applicability
and practicability of the guideline
and the defined set of KPIs
© Omnitele Ltd. 2007 10
KPI Name
Paging failure rate
Paging Intensity
RRC connection setup success rate
RAB setup success rate
Setup success rate
RRC connection blocking rate
RAB blocking rate
Congestion
Cell downtime
RRC connection drop rate
RAB drop rate
Drop due to system rate
Active set update success rate
Soft handover overhead
Outgoing CS inter-RAT handover success rate
Incoming CS inter-RAT handover success rate
Outgoing PS inter-RAT handover success rate
Incoming PS inter-RAT handover success rate
Inter-Frequency handover success rate
Hard HSDPA cell change success rate
Drop rate for drops due missing neighbours
UL DCH BLER before combining
UL DCH BLER after combining
UL DPDCH BER
Retransmission rate
Total CS traffic
Total payload traffic
Throughput
Integrity
Spreading factor usage
Code allocation failure rate
Number of users
Total downlink TX power
Downlink TX power
Downlink transmitted code power
Average uplink RSSI
Uplink RSSI
Channel switching success rate
Downswitching due to system
Compressed mode users
Compressed mode start to handover rate
 Case Study

• A case study was conducted in live network

• Purpose was to test the KPIs in practise
• The number of KPIs was limited due to operator
• Also field measurements was performed in order to find the
reasons for the poor performance, due to limited set of KPIs
• The performance of the network was improved

© Omnitele Ltd. 2007 11

 Case study:High dropped call rate
350 35,00

300 30,00

• Worst performing cells 250 25,00

• Cell A was the worst 200 20,00

Number of Dropped Calls (left)

[%]
Dropped Call Rate (right)
performing cell 150 15,00

100 10,00

50 5,00
25 80,00

70,00
0 0,00
20 Cell A Cell B Cell C Cell D Cell E Cell F Cell G Cell H Cell I Cell J
60,00

50,00
15

Number of Dropped Calls (left)

[%]

40,00
Dropped Call Rate (right)
10
30,00

20,00
5

10,00

0 0,00
2.

2.

2.

2.

2.

2.

2.

2.
.

.
12

12

12

12

12

12

.1

.1

.1

.1

.1

.1

.1

.1
4.

5.

6.

7.

8.

9.

10

11

12

13

14

15

16

17

Date

© Omnitele Ltd. 2007 12

 Case study:High dropped call rate

Solution:
• Most common drop reasons are incorrect HO parameters, interference or
lack of coverage
• Handover KPIs were checked, but the KPIs were ok→ no incorrect HO
parameters
• Since the cell was located in the city area, the coverage was assumed to
be ok. The high number of drops indicated that the drops was not caused
by lack of indoor coverage.
• Interference was the most probable cause. Since the KPI set was limited,
field measurements were performed.
• Field measurements showed that there was a neighbour relation missing
between Cell A and a another cell.
– Every cell that is not defined as neighbour to the serving cell, is seen as
interference by the mobile.
– If the signal strength of the other cell is strong enough, the call is disconnected
by the network. The call can also drop due to interference caused by the
interfering cell.
• This problem would been possible to detect with the KPI defined for drops
due to missing neighbours.

© Omnitele Ltd. 2007 13

 Case study:High dropped call rate

• After the reason for high

dropped call rate was 20 40,00

18
found, the missing 16
35,00

neighbour relation was 14

30,00

defined 12
25,00

Number of Dropped Calls (left)

[% ]
10 20,00
• The change improved 8
Dropped Call Rate (right)

the performance of the 6

15,00

Cell A greatly and also 4

10,00

the performance of the 2

5,00

other cell was improved 0

1.1. 2.1. 3.1. 4.1. 5.1. 6.1. 7.1. 8.1. 9.1. 10.1. 11.1. 12.1. 13.1. 14.1.
0,00

Date

• There are no dropped

calls in Cell A after the
change

© Omnitele Ltd. 2007 14

 Conclusions

• Monitoring and optimisation is important part of operating and

maintaining WCDMA networks
• Increased traffic rises new problems for the operators
• Right KPIs and proper use of the KPIs will help to maintain and
improve the performance of WCDMA networks
• Case study showed that the KPIs and the guideline are
applicable to be used in practise, however the case study was
limited to basic KPIs that the operator had. The use of whole
KPI set will give more information on the performance of the
WCDMA radio network

© Omnitele Ltd. 2007 15

 Future research

• Performance measurement and KPI standards are quite

immature yet, more work needs to be done
• The implementation of counters vary between different vendors,
the same KPI definitions are not applicable in different networks
• The set of counters are not as mature as for example in GSM
networks. More counters are needed to be able to define more
KPIs
• New network features (HSPA) need new counters and new
KPIs. The development of the KPIs is needed constantly

© Omnitele Ltd. 2007 16